The Gemini API has expanded its File Search tool to support retrieval-augmented generation (RAG) systems that can handle multimodal data and custom metadata. This enhancement allows users to process text and images together, improving the organization and searchability of visual assets. The updated tool, powered by the Gemini Embedding 2 model, enables applications to search for images based on emotional tone or visual style described in natural language. Additionally, it introduces page citations for better grounding, allowing users to verify the source of information accurately. The integration of custom metadata filters enhances the precision of searches, significantly reducing irrelevant results and improving RAG workflows.
RAGMultimodalNLP+1
RAGMultimodalNLPGenerative AI
Gemini API Enhances File Search with Multimodal Capabilities
Hacker News· gmays· Monday, May 11, 2026
More Articles From This Day
NVIDIAOpenAI+2
NVIDIAOpenAIAI InvestmentsCompute Capacity
NVIDIA Commits Over $40 Billion to AI Equity Investments, Highlighting Major OpenAI Stake
NVIDIA has allocated more than $40 billion to AI equity investments within the first four months of 2026, as reported by CNBC based on public filings. The largest single investment of $30 billion was directed to OpenAI, while the remaining funds are distributed among several companies including CoreWeave, IREN, and Corning. NVIDIA's strategy appears to focus on vertical integration rather than traditional venture capital, with investments structured as warrants or commitments rather than outright equity. These investments aim to bolster NVIDIA’s influence in the AI value chain, ensuring compute capacity around its hardware. NVIDIA’s CFO Colette Kress emphasized that this investment strategy signals the company's upstream and downstream positioning within the AI sector.